Algorithm Design Using EMG To Control Limb Motor Function Rebuilding Between Different Limbs And Its Hardware Implementation

In China, there are over 80 million people suffering from all types of disabilities and among them 24 million are physically disabled. A great number of these physical disabilities are caused by spinal cord injury or stroke. However, the pathogenesis of the two diseases are different:The spinal cord injury is caused by the damage of the pathway from brain to body, so that the motor and sensory neural signals can not be transferred normally, thus the paraplegic is resulted. Stroke patients’ pathways of neural signals are intact, but their brains are barely able or unable to produce movement controlling signals, thus the hemiplagic is resulted. So researches on the therapies of paraplegia and hemiplegia have significant scientific significance and clinical application. In this case, "embedded microelectronic neural channel bridging SOC" has been proposed by our research group. Implantable nerve electrodes are in need of surgical implantation, which causes often infection. Therefore, the leaders of our researching group have brought a new idea forward, using the surface EMG (electromyography) as a source or reference signal of surface FES (Functional Electrical Stimulation).This paper mainly carrys out the study of EMG recognition pattern of human limb function rebuilding system based on microelectronics-embedded channel bridging, finally realized the real-time control between two different healthy people to realyze four kinds of actions:grasping, wrist extension, wrist flexion, and finger extension.First of all, through the comparison of recognition rate and calculation rate among time domain features, auto regression model coefficients and sample entropy, the time domain features are chosen as feature parameters. And then the linear discriminant analysis classification algorithm is introduced to realize the recognition of four different actions and the influence of different force of training data on the recognition rate is studied.Secondly, the stimulating pulses are generated by the recognition of the spikes, for which the threshold detection method and the slope sign change detection method are used. Raster graphics are obtained by detecting EMG potential spikes. Each raster graphic representes an EMG potential spike according to the negative stimulation pulses are also generated. A comparison of two methods is also performed.Then, the embedded implementation of EMG signal control algorithm in microelectronics neural-muscle bridging system is studied. The EMG recognition algorithm is implanted into STM32F407 and stimulating pulses generation algorithm is into STM32F103.Finally, the combination experiment is performed in eight channels microelectronic neural-muscle bridging system and verified in healthy limb motor function rebuilding experiments, to realize that one healthy person drives another one to finish grasping, wrist extension, wrist flexion and finger extension in real time.